Identification of Developmentally Expressed Proteins That Functionally ...

19 downloads 0 Views 929KB Size Report
Oct 17, 2001 - to specific subnuclear locations (7). In addition, protein modi- fication by the ubiquitin-like proteins SUMO and Nedd8 has been found to have a ...
THE JOURNAL

OF

BIOLOGICAL CHEMISTRY

Vol. 277, No. 4, Issue of January 25, pp. 2897–2907, 2002 Printed in U.S.A.

Identification of Developmentally Expressed Proteins That Functionally Interact with Nedd4 Ubiquitin Ligase* Received for publication, October 17, 2001 Published, JBC Papers in Press, November 20, 2001, DOI 10.1074/jbc.M110047200

Rodolfo Murillas‡, Kimberly S. Simms‡, Shigetsugu Hatakeyama§¶, Allan M. Weissman§, and Michael R. Kuehn‡储 From the ‡Experimental Immunology Branch, NCI, National Institutes of Health, Bethesda, Maryland 20892-1360 and the §Regulation of Protein Function Laboratory, NCI, National Institutes of Health, Bethesda, Maryland 20892-1152

Nedd4 is a HECT domain-containing ubiquitin ligase that mediates ubiquitylation and proteasome degradation of target proteins. The molecular basis for the interaction of Nedd4 with substrates lies in its WW domains, which can bind proline-rich (PY) domains in target proteins. Nedd4 is a developmentally expressed protein and may have a fundamental role to play in embryonic processes. However, whether Nedd4 has such a function is currently unknown, in part because few developmentally regulated ubiquitylation substrates have been identified or characterized. We have carried out a yeast two-hybrid screen and identified four proteins expressed in the mid-gestation embryo that are able to interact with Nedd4. Characterization of their functional interaction with Nedd4 in vitro and in vivo demonstrated that three of the four are bona fide Nedd4 binding partners, and two have the capacity to be ubiquitylation substrates. One of these is the first identified nonviral substrate for Nedd4-mediated monoubiquitylation. Interestingly, neither of these two ubiquitylated proteins interacts with Nedd4 through PYmediated mechanisms. For one of the three Nedd4 binding partners, there was no discernable evidence of ubiquitylation. However, this protein clearly associates with Nedd4 through its PY domains and can alter the location of Nedd4 in cells, suggesting a role other than as a ubiquitylation substrate.

A large number of recent findings have highlighted the importance of protein modification by the covalent addition of small polypeptides (1, 2). The prototype for this is ubiquitin, a 76-amino acid polypeptide that, when added in multiple units to form a chain (polyubiquitylation), targets proteins for degradation by the 26 S proteasome (3). In contrast, the addition of a single ubiquitin moiety (monoubiquitylation) has been implicated in receptor endocytosis (4, 5), activation of gene expression through histone modification (6), and targeting of proteins to specific subnuclear locations (7). In addition, protein modification by the ubiquitin-like proteins SUMO and Nedd8 has been found to have a wide range of functions, from control of subcellular localization to modulation of protein stability (8, 9).

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ¶ Current address: Dept. of Molecular and Cellular Biology, Medical Institute of Bioregulation, Kyushu University, Fukuoka 812-8582, Japan. 储 To whom correspondence should be addressed: Bldg. 10/Rm. 4B-36, 10 Center Dr., Bethesda, MD 20892-1360. Tel.: 301-435-6476; Fax: 301-496-0887; E-mail: [email protected]. This paper is available on line at http://www.jbc.org

Ubiquitylation occurs at lysine residues, either within target proteins or on ubiquitin already attached to the target protein, and involves three distinct enzymatic activities. Ubiquitin-activating enzyme (E1)1 activates ubiquitin to a high energy state in an ATP-dependent manner and then transfers it to a ubiquitin-conjugating enzyme (E2). The E2 either can transfer ubiquitin directly to the target substrate or interact with a ubiquitin protein ligase (E3), which then mediates the transfer of ubiquitin to substrate proteins (10). There are two major classes of E3 ubiquitin ligases. Those containing a RING finger motif act in concert with E2 enzymes in the direct transfer of ubiquitin from the E2 to the target protein (11). Members of the second class of E3 ubiquitin ligases accept activated ubiquitin from the E2 and transfer it to substrate proteins. The first identified member of this class is E6-AP, which mediates polyubiquitylation of p53 in conjunction with human papillomavirus E6 protein (12, 13). Other members of this family include Nedd4 and related Nedd4-like proteins (14). These all have a carboxyl-terminal region known as the HECT (for homologous to E6-AP carboxyl terminus) domain, which provides the ubiquitin ligase enzymatic activity. Unlike E6-AP, members of the Nedd4 family have two to four tryptophan-based WW domains, which bind certain proteins containing proline-rich motifs (Fig. 1A). The WW domains of Nedd4 have been categorized as group I, which bind preferentially to the consensus sequence PPXY (PY domain) (15). Group II WW domains typically bind to the sequence PPLP (16), whereas a third class is able to bind other proline-rich consensus sequences (17). The fourth class of WW domains binds in a proline-independent manner, depending instead on phosphoserine and/or phosphothreonine residues (18). The best-characterized Nedd4 substrate is the epithelial sodium channel, a plasma membrane protein composed of several subunits, two of which are ubiquitylated by Nedd4. The WW domains of Nedd4 bind to PY domains in the cytoplasmic region of these proteins. Truncation of the PY domains results in a reduced turnover rate of epithelial sodium channel and leads to Liddle’s syndrome, an inherited form of hypertension (19, 20). Nedd4 was identified originally as a gene expressed at high levels in neural precursor cells during development and subsequently down-regulated in the adult (21). Nedd4 also is expressed at high levels in proliferating chondrocytes in midgestation embryos (22, 23). These sites of expression suggest an important role for Nedd4 in regulating developmentally impor-

1 The abbreviations used are: E1, ubiquitin-activating enzyme; E2, ubiquitin carrier protein; E3, ubiquitin-protein isopeptide ligase; EST, expressed sequence tag; TBS, Tris-buffered saline; TBST, Tris-buffered saline plus Tween 20; ␤-gal, ␤-galactosidase; GST, glutathione S-transferase; BSA, bovine serum albumin; ATP␥S, adenosine 5⬘-O-(thiotriphosphate); PBS, phosphate-buffered saline; HA, hemagglutinin.

2897

2898

Nedd4 Developmental Targets

tant gene function at the level of protein half-life. Indeed, the actions of two Nedd4-like proteins, Suppressor of Deltex and Smurf1, are consistent with such a role during development. Smurf1 mediates polyubiquitylation of Smad1, a mediator of bone morphogenetic protein signaling (24), and Suppressor of Deltex regulates Drosophila Notch signaling (25). The Nedd4 family member, Itch, also can ubiquitylate Notch (26). However, the specific functions of Nedd4 itself in developmental processes are unknown, in part because of the fact that very few developmentally expressed ubiquitylation substrates have been characterized (27, 28), although a number of potential binding partners have been identified (23). Therefore, we searched for proteins expressed in the mid-gestation embryo that interact with Nedd4 in a yeast two-hybrid screen, and further characterized these to determine their functional interaction with Nedd4 both in vitro and in vivo. We have identified four proteins, which we designated as Nedd4 binding partners 1– 4 (N4BP1– 4). We demonstrate here that N4BP1 and N4BP2, both novel proteins, have the capacity to be ubiquitylation substrates. N4BP1 is a bona fide substrate for Nedd4 ubiquitin ligase activity in vivo, and the first identified nonviral target for Nedd4-mediated monoubiquitylation. N4BP3, also a novel protein, is not a ubiquitylation substrate, but it can alter the subcellular location of Nedd4, indicating a functional interaction. EXPERIMENTAL PROCEDURES

Yeast Two-hybrid Screening—A fragment of mouse Nedd4 cDNA encoding amino acids 199 –777 cloned into the GAL4 binding domain vector pGBT9, and an 11-day mouse embryo cDNA library cloned into the GAL4 activation domain vector pGAD10 were sequentially used to transform Hf7c yeast cells according to the MATCHMAKER two-hybrid system protocol (CLONTECH, Palo Alto, CA). Transformants were plated on selection media (lacking tryptophan, leucine, and histidine) containing 25 mM 3-aminotriazole. After incubating 10 days at 30 °C, clones expressing HIS3 and ␤-galactosidase (␤-gal) were identified. Plasmids were recovered according to MATCHMAKER protocols, and transformed into yeast containing either pGBT9-Nedd4 or the pGBT9 empty vector. Inserts from plasmids that activated GAL4 transcription with pGBT9-Nedd4 were excised from pGAD10 and subcloned into pcDNA3.1 (Invitrogen, Carlsbad, CA), in frame with the Xpress and 6-histidine (His) epitope tags. Plasmid Construction and in Vitro Mutagenesis—A full-length N4BP1 expression vector was made in pcDNA3 (Invitrogen) by assembling the original clone with mouse EST AA444325 and a fragment containing the 5⬘ part of the coding region, produced by PCR using a N4BP1 genomic cosmid clone as template. KIAA0341, a human fulllength cDNA homologous to N4BP3, was obtained from Kazusa DNA Research Institute as a pBluescript clone. The insert was removed and subcloned into pCDNA3.1. Site-directed mutagenesis of the proline-rich sites in full-length mouse N4BP1 and KIAA0341 was done using QuikChange (Stratagene, La Jolla, CA). Recombinant Protein Expression—All Nedd4 GST fusion constructs used in this study were described previously (29). They were generated using the pGEX GST Gene Fusion System (Amersham Biosciences, Inc.) and expressed in Escherichia coli BL21 (Novagen, Madison, WI). The E1 from wheat cloned in pET3a and human UbcH5B cloned in pET15b were expressed in E. coli BL21 using the pET expression system (Novagen) as described previously (29). In vitro expression and radiolabeling of proteins was performed by coupled transcription and translation using the TnT wheat germ extract kit (Promega, Madison, WI), in the presence of [35S]methionine (Amersham Biosciences, Inc.) according to the manufacturer’s protocol. To produce antisera against N4BP1 (anti-N4BP1), the insert was subcloned from pGAD10 into the EcoRI site of pGEX-4T-1 (Amersham Biosciences, Inc.) and expressed in BL21 cells. Recombinant fusion protein was purified by glutathioneSepharose, and ⬃100 ␮g was injected into rabbits at 3-week intervals. In Vitro Binding—Nedd4 GST fusion proteins (GST-Nedd4, GSTNedd4:N, and GST-Nedd4:C) were adsorbed to glutathione-Sepharose beads and then combined with each N4BP protein or E6-AP, which had been in vitro translated and 35S-labeled, in binding buffer (150 mM NaCl, 20 mM Tris-HCl, pH 7.4, 5 mM dithiothreitol, 0.5% Nonidet P-40) with 2 mg/ml BSA. Mixtures were incubated 20 h at 4 °C with agitation and then washed four times with five volumes of binding buffer to

remove unbound proteins. Bound protein was eluted by boiling in reducing sample buffer, and then analyzed by SDS-PAGE and autoradiography. In Vitro Ubiquitylation—N4BP proteins, in vitro translated and radiolabeled as above, were incubated in the presence of GST-Nedd4, BL21 bacterial extracts of wheat E1and human UbcH5b, and ubiquitin, in 25 mM Tris-HCl, pH 7.6, 120 mM NaCl, 3 mM dithiothreitol, 1 mM MgCl2, 1 mM phosphocreatine, 100 units of creatine phosphokinase, 0.6 units/ml inorganic pyrophosphatase, and 5 mM ATP␥S (Roche Molecular Biochemicals) for 1 h at 30 °C. Reactions were terminated with the addition of SDS-containing reducing sample buffer, resolved by SDSPAGE, and visualized by autoradiography. Cell Culture, Transfection, Western Blotting, and Immunoprecipitation—HEK293 cells were cultured in Dulbecco’s modified minimal essential medium (Invitrogen) supplemented with 10% fetal bovine serum. Transfections were made using the calcium phosphate method. After 24 – 48 h, cells were lysed in buffer containing 50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.5% Triton X-100, 10 ␮g/ml aprotinin, 10 ␮g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride. The insoluble fraction was removed by centrifugation, and the cleared supernatant was either analyzed directly by Western blotting or used for immunoprecipitations. For N4BP1 immunoprecipitation, lysates were incubated for 16 h at 4 °C with anti-N4BP1 polyclonal antisera. For N4BP2 and N4BP3 immunoprecipitation, lysates were incubated with anti-Xpress or anti-HisG monoclonal antibody (Invitrogen). Immunoprecipitates were collected by incubating with protein A/G-agarose (Santa Cruz Biotechnology, Santa Cruz, CA) for an additional 1 h. After brief centrifugation, the complexes were washed three times with icecold 50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, and 0.1% Triton X-100. Proteins were recovered by boiling in SDS sample buffer, fractionated by SDS-PAGE, and transferred to polyvinylidene difluoride membranes (Novex, San Diego, CA). Membranes were blocked in 5% milk in TBS and incubated for 2 h with the different primary antibodies diluted in TBS. Antibody dilutions used were: anti-N4BP1, 1:2000 dilution; anti-Nedd4, 1:1000; anti-Xpress and anti-HisG, 1:2000; and anti-Myc monoclonal antibody (Invitrogen), 1:1000. After incubation with primary antibodies, membranes were washed with TBST and incubated for 1 h with the appropriate peroxidase-coupled secondary antibodies (Pierce). After washing with TBST, chemiluminescent detection was performed using Supersignal (Pierce). For proteasome inhibition, stock solutions of MG101 (calpain inhibitor I), lactacystin, MG132, and epoxomycin (Calbiochem, San Diego, CA) were prepared in Me2SO and diluted immediately before use in Dulbecco’s modified Eagle’s medium to final concentrations of 100 ␮M for MG101, 5 ␮M for lactacystin, 10 ␮M for MG132, and 5 ␮M for epoxomycin. Subcellular Localization—HEK293 cells were plated on glass coverslips, transiently transfected, and, 24 h later, fixed in 4% paraformaldehyde and permeabilized with 0.2% Triton X-100 in PBS. Cells were then washed with 0.1% BSA in PBS, incubated for 2 h with primary antibodies diluted in 0.1% BSA in PBS, washed in PBS, and further incubated with fluorescein isothiocyanate-conjugated anti-rabbit antibody (Vector Laboratories, Burlingame, CA) or rhodamine-conjugated anti-mouse antibody (Pierce) for 1 h. N4BP1 was detected with antiN4BP1 antiserum used at 1:400 dilution. His-tagged N4BP2 and N4BP3 were detected with anti-HisG monoclonal antibody used at 1:5000 dilution. HA-tagged Nedd4 was detected with either anti-HA monoclonal antibody (Covance Research Products, Denver, PA) at 1:1000 dilution or anti-Nedd4 polyclonal antiserum at 1:1000 dilution. After incubation with secondary antibodies, samples were mounted in ProLong (Molecular Probes, Eugene, OR) plus 4⬘,6-diamidino-2-phenylindole, observed with a Zeiss Axioplan fluorescence microscope, and imaged with a SPOT2 digital camera. RESULTS

Isolation of Developmentally Expressed Nedd4 Interacting Proteins—To identify potential Nedd4 ubiquitylation substrates that might play a role in embryo development, a fragment of Nedd4 containing WW domains 2 and 3 and the HECT domain (Fig. 1B) was used as bait in a yeast two-hybrid screen of a mid-gestation mouse embryo cDNA library. From 7 ⫻ 106 clones screened, four interacting clones were identified. These were designated N4BP1– 4 (Fig. 2). N4BP4 encodes an aminoterminally truncated form of PLIC-2 (Fig. 2D), a protein recently characterized as a regulator of the interaction between the proteasome and ubiquitin ligases (30). N4BP1–3 show no similarity to proteins with known functions. Taking advantage

Nedd4 Developmental Targets

2899

FIG. 1. Structure of Nedd4 and Nedd4 constructs used in this study. A, full-length Nedd4 protein containing the amino-terminal C2 domain, three WW domains, and the carboxyl-terminal HECT domain. B, an amino-terminally truncated form of Nedd4, extending from amino acid position 199 to the carboxyl terminus, used for yeast two-hybrid screening. C, an amino-terminally truncated form of Nedd4, extending from amino acid position 52 to the carboxyl terminus, used as a nearly complete Nedd4 for in vitro binding (GST pulldown experiments). D, the amino-terminal half of Nedd4 used for in vitro binding, extended from amino acid position 52 to 422. E, the carboxyl-terminal half of Nedd4 used for in vitro binding, extended from amino acid position 423 to the carboxyl terminus.

of overlapping ESTs and, in the case of N4BP1, a genomic clone that we isolated covering the 5⬘ region of the gene, the fulllength open reading frames encoded by N4BP1 and N4BP3 were deduced (Fig. 2, A and C). For clarity these are referred to as flN4BP1 and flN4BP3, respectively. For N4BP1–3, amino acid sequences of full-length human orthologs were found or deduced. For flN4BP1, the 893-amino acid human protein KIAA0615 was found to have 89% similarity. N4PB2 was found to be 92% similar to the carboxyl region of a full-length open reading frame encoding 1770 amino acids assembled from KIAA1413 and EST AK001542. flN4BP3 was found to be 95% similar to KIAA0341. N4BP1–3 Bind Nedd4 in Vitro—To confirm the yeast twohybrid results, N4BP1– 4 were in vitro translated and radiola-

beled with [35S]methionine. These were then evaluated for binding to glutathione-Sepharose-immobilized bacterially expressed GST-Nedd4 fusion proteins that included either both the WW and HECT domains (GST-Nedd4; Fig. 1C), the WW domain-containing region only (GST-Nedd4:N; Fig. 1D) or the HECT domain only (GST-Nedd4:C; Fig. 1E). N4BP1 bound to both GST-Nedd4 (Fig. 3A, lane 2) and GST-Nedd4:N (Fig. 3A, lane 3). There was little detectable binding to GST-Nedd4:C (Fig. 3A, lane 4) or to a non-WW domain HECT E3, E6AP, which was used as an additional negative control (Fig. 3A, lane 6). Because regions within the amino-terminal half of Nedd4 were used in the two-hybrid screen, in vitro binding to GSTNedd4:N serves to validate the interaction in yeast. Similar results were found for N4BP2 (Fig. 3B) and for N4BP3 and its

2900

Nedd4 Developmental Targets

FIG. 2. N4BP proteins identified in this study. The four N4BP proteins are shown aligned with ESTs or known gene sequences with which they exhibit significant homology. A, the original N4BP1 isolate contains the carboxyl-terminal 685 amino acids of the 893-amino acid full-length mouse protein (flN4BP1), which is homologous to human KIAA0615. flN4BP1 was constructed using EST AA444325 and a genomic fragment. The proline-rich domains in N4BP1 (thick black lines) are not conserved in KIAA0615. B, the 374-amino acid original isolate of N4BP2 is homologous to the carboxyl region of a 1770-amino acid human protein (flhN4BP2), deduced from KIAA1413 and EST AK001542. C, the original N4BP3 isolate contains the middle 220 amino acids of the 537-amino acid full-length mouse protein (flN4BP3), which is homologous to human KIAA0341. The proline-rich domains in N4BP3 (thick black lines) are conserved in KIAA0341. D, the original isolate of N4BP4 contains the carboxyl-terminal 189 amino acids of the 638-amino acid protein, Plic2. The ability (⫹) or inability (⫺) of specific N4BP proteins to bind or to be ubiquitylated by Nedd4, either in vitro or in vivo, is indicated in the columns to the right. nd indicates not determined.

Nedd4 Developmental Targets

2901

FIG. 3. In vitro binding of N4BP proteins to Nedd4. N4BP proteins were in vitro translated with [35S]methionine and incubated with bacterially produced GST-Nedd4, GST-Nedd4:N, GST-Nedd4:C, GST, or GST-E6AP proteins bound to glutathione-Sepharose. Protein complexes were collected by centrifugation, washed, and analyzed by SDS-PAGE and autoradiography. 20% of the amount of each in vitro translated protein used for binding (labeled 20% IVT) was loaded as a control. A–C, N4BP1 (A), N4BP2 (B), and N4BP3 and KIAA0341 (C) each bind to GST-Nedd4 (lane 2 in each panel) and GST-Nedd4:N (lane 3 in each panel) equally well, but show minimal binding to GST-Nedd4:C (lane 4 in each panel) and GST-E6AP (lane 6 in each panel) and no binding to GST (lane 5 in each panel). D, N4BP4 shows no binding to Nedd4. E, N4BP1 carrying mutations in the PY motif (lane 6), in both PPLP motifs (lane 8), or in the PY and PPLP motifs (lane 10) binds Nedd4 identically to wild-type N4BP1 (lane 3). F, binding to Nedd4 by KIAA0341 carrying a mutated PY motif (lane 4) is greatly reduced compared with wild-type KIAA0341 (lane 2).

full-length human counterpart KIAA0341 (Fig. 3C). However, no binding to N4BP4 (Plic-2 carboxyl terminus) was observed. Therefore, no further studies were carried out with this potential binding partner. Nedd4 can bind other proteins through the interaction of its WW domains with proline-rich domains in target proteins. Thus, identification of proline-rich regions in N4BP proteins may provide clues for their interaction with Nedd4. In fact,

N4BP1 contains a PY motif (PPEY) and two PPLP motifs. Strikingly, however, none of these three regions are conserved in the human ortholog (KIAA0615). N4BP3 also contains a consensus PY motif (PPPY), and this sequence is conserved in the corresponding region of KIAA0341. In the case of N4BP2, neither this isolate nor the corresponding region of the human ortholog has an identifiable WW interaction domain. N4BP4 (PLIC-2) also lacks proline-rich regions. The lack of conserva-

2902

Nedd4 Developmental Targets

FIG. 4. In vitro ubiquitylation of N4BP1–3. N4BP proteins were in vitro translated with [35S]methionine and incubated in a ubiquitylation reaction mix containing E1 activity, and with or without GST-Nedd4, E2 (UbcH5), and ubiquitin as indicated, followed by SDS-PAGE and autoradiography. A, in the presence of all ubiquitylation components (lane 2), N4BP1 shows only three higher molecular weight bands (bracketed), indicating limited ubiquitylation. In the absence of any component (lanes 1 and 3–5), there is no ubiquitylation. B, in the presence of all ubiquitylation components (lane 2), N4BP2 shows a smear of higher molecular weight bands suggesting polyubiquitylation. No higher molecular weight species are seen in the absence of critical ubiquitylation reaction components (lanes 1 and 3–5). C, human N4BP3 (KIAA0341) is not in vitro ubiquitylated.

tion of the proline-rich regions between N4BP1 and its human counterpart raises the possibility that these domains are not required for the interaction with Nedd4. To address this, key residues in the PY domain and two PPLP domains (PPLP1 and PPLP2; Fig. 2A) of N4BP1 were mutated. As shown in Fig. 3E, these mutations had no discernable effect on N4BP1 binding to Nedd4. Thus the interaction between Nedd4 and N4BP1 is, in fact, independent of these proline-rich regions. Conversely, mutation of the evolutionarily conserved PY domain in the human ortholog of N4BP3 (KIAA0341) dramatically decreased Nedd4 binding (Fig. 3F), establishing that the binding of this protein to Nedd4 is dependent on interactions between the Nedd4 WW domains and the proline-rich region of N4BP3. N4BP1 and N4BP2 Are in Vitro Ubiquitylated by Nedd4 —As N4BP1–3 all bind Nedd4, we next addressed their capacity to serve as in vitro substrates for this E3. In the presence of GST-Nedd4, ubiquitin, and E1 and E2 enzyme activities, N4BP1 was ubiquitylated as indicated by the ubiquitin and E2-dependent appearance of discrete higher molecular weight species (Fig. 4A, lane 2, bracketed). Under the same conditions N4BP2 underwent what appeared to be polyubiquitylation, as demonstrated by the appearance of a smear of higher molecular weight bands that extended up the gel accompanied by a significant decrease in the amount of the non-ubiquitylated form of this protein (Fig. 4B, lane 2, bracketed). Again, this was dependent on the presence of E1, E2, and ubiquitin (Fig. 4, A and B, lanes 1, 3, 4, and 5). In contrast, full-length human N4BP3 (KIAA0341) was not ubiquitylated despite its clear capacity to bind Nedd4 (Fig. 4C, lane 2). N4BP2 and N4BP3 Bind Nedd4 in Vivo—To extend the in vitro observations to cells, eukaryotic expression vectors encoding flN4BP1, N4BP2, or the full-length human N4BP3 (KIAA0341) were co-transfected with Nedd4 into HEK293 cells and evaluated for co-immunoprecipitation with Nedd4. Immunoprecipitation of N4BP1 was performed using antiserum derived against bacterially expressed protein (anti-N4BP1). N4BP2 and KIAA0341 were immunoprecipitated using mono-

clonal antibodies directed against their amino-terminal His and Xpress epitope tags. No co-immunoprecipitation of Nedd4 with N4BP1 was detected (Fig. 5A, lanes 1–3), despite clear evidence for expression of this protein (Fig. 5B, lanes 1–3). In contrast, in vivo association of N4BP2 and KIAA0341 could be discerned readily (Fig. 5A, lanes 5 and 7). Although there was less Nedd4 co-immunoprecipitated with N4BP2 than with KIAA0341, the expression level achieved in the transfection was much lower for N4BP2 than for KIAA0341 (compare lanes 5 and 7 in Fig. 5B). Thus, the level of interaction with Nedd4 may be similar for these two proteins. Consistent with the in vitro binding data, mutation of the PY domain in KIAA0341 resulted in a dramatic decrease of co-immunoprecipitated Nedd4 (Fig. 5A, lane 9), underscoring the importance of the PY domain for interactions between this protein and Nedd4. N4BP1 Is Monoubiquitylated in Vivo by Nedd4 —The failure to see co-immunoprecipitation of Nedd4 and N4BP1 does not preclude a functionally significant in vivo interaction. To evaluate this, N4BP1 ubiquitylation was analyzed in HEK293 cells. When we co-transfected flN4BP1 with a Myc epitope-tagged ubiquitin expression vector (31), a single Myc-ubiquitin immunoreactive band could be detected after immunoprecipitation with anti-N4BP1 (Fig. 6A, lane 3). Its mobility was consistent with its being a monoubiquitylated form of flN4BP1. Strikingly, co-transfection with a Nedd4 expression vector resulted in a marked increase in monoubiquitylated flN4BP1 (Fig. 6A, lane 4), despite the fact that the total N4BP1 protein was equivalent in all lanes, as assessed by immunoblotting with anti-N4BP1 (Fig. 6B, lanes 1– 4). Thus, these results indicate a functional interaction between Nedd4 and N4BP1 in vivo. N4BP2 Is Polyubiquitylated in Vivo—To assess in vivo ubiquitylation of N4BP2 and N4BP3, His- and Xpress-tagged forms of N4BP2 and N4BP3 were co-expressed with Myc-tagged ubiquitin in HEK293 cells. Following immunoprecipitation with anti-His antibody and Western blotting with anti-Myc antibody, multiple high molecular weight forms could easily be discerned for N4BP2 (Fig. 7A, lanes 1 and 2 showing duplicate

Nedd4 Developmental Targets

2903

FIG. 5. In vivo binding of N4BP1–3 to Nedd4. Expression vectors for flN4BP1, N4BP2, and flhN4BP3 (KIAA0341) were transfected into HEK293 cells, with or without Nedd4 overexpression. A, cell lysates were immunoprecipitated using anti-N4BP1 antiserum (for N4BP1) or anti-His monoclonal (for N4BP2 and N4BP3), followed by immunoblotting with anti-Nedd4 antiserum. The arrow indicates the position of co-immunoprecipitated Nedd4 found for N4BP2 (lane 5) and KIAA0341 (lane 7). No co-immunoprecipitated Nedd4 was found for N4BP1 (lane 2). A greatly reduced amount of co-immunoprecipitated Nedd4 was found for the PY mutated N4BP3 (lane 9). B, the amount of each immunoprecipitated N4BP protein was determined by reblotting with antiN4BP1 antiserum for N4BP1 and antiXpress monoclonal antibody for N4BP2 and N4BP3. Asterisks mark the position of the bands corresponding to each immunoprecipitated protein. C, the amount of Nedd4 expression in each sample was determined by immunoblotting whole cell lysates with anti-Nedd4 antiserum.

samples). In contrast, no such forms were detected for N4BP3 (data not shown). To assess whether Nedd4 mediates N4BP2 polyubiquitylation in vivo, HEK293 cells were co-transfected with Nedd4, N4BP2, and Myc-tagged ubiquitin expression vectors. No discernable change in N4BP2 polyubiquitylation was detected (Fig. 7A, lanes 3 and 4 showing duplicate samples) despite clear overexpression of Nedd4 (Fig. 7C, lanes 3 and 4) compared with control transfections (Fig. 7C, lanes 1 and 2). Equal loading was verified by reblotting with anti-Xpress antibody (Fig. 7B, lanes 1 and 4). Overexpression of Nedd4 also did not result in any detectable ubiquitylation of N4BP3 (data not shown). These data correlate with the in vitro ubiquitylation results, in which N4BP2 was polyubiquitylated and N4BP3 was not ubiquitylated at all. However, there is no evidence that Nedd4 mediates N4BP3 polyubiquitylation in vivo. Proteasome Inhibition Increases the Level of N4BP2 but Not

N4BP1—Because proteins targeted for proteasomal degradation are generally modified with chains of four or more ubiquitins, we reasoned that N4BP2 but not N4BP1 might be degraded by the proteasome. To assess this, we determined the effect of proteasome inhibitors on steady state levels of N4BP1 and N4BP2 expressed in HEK293 cells (Fig. 8). The steady state level of N4BP1, as determined by immunoblotting of whole cell lysates with anti-N4BP1 antiserum, was not affected by any treatment (Fig. 8A). In contrast, following treatment with different proteasome inhibitors and immunoprecipitation and immunoblotting with anti-Xpress, a significant increase in N4BP2 steady state levels was found (Fig. 8B, bands marked by asterisk). Notably, levels of co-transfected Xpress-tagged ␤-gal, a stable protein insensitive to proteasome inhibition used as a control, were unaffected (Fig. 8B, arrow). Thus N4BP2, which is modified by multiple ubiquitin moieties both in cells and in vitro, is subject to proteasome-mediated degra-

2904

Nedd4 Developmental Targets

FIG. 6. In vivo monoubiquitylation of N4BP1 by Nedd4. HEK293 cells were transfected with expression vectors for either N4BP1 or flN4BP1, with or without Nedd4 overexpression, in the presence of Myc-tagged ubiquitin. N4BP1 and flN4BP1were immunoprecipitated from cell lysates using anti-N4BP1 antiserum. A, immunoblotting with anti-Myc monoclonal antibody reveals a single band corresponding to a monoubiquitylated form of flN4BP1 (lane 3). The amount of monoubiquitylated flN4BP1 is greatly enhanced by Nedd4 overexpression (lane 4). The truncated form of N4BP1 is not efficiently ubiquitylated (lanes 1 and 2). B, an equivalent amount of N4BP1 or flN4BP1 protein was expressed (lanes 1 and 2 or lanes 3 and 4) as determined by reblotting with anti-N4BP1 antiserum. C, significant Nedd4 overexpression was achieved (lanes 2, 4, and 5), as determined by analyzing whole cell lysates by immunoblotting with anti-Nedd4 antiserum.

dation, whereas N4BP1, which demonstrates a low stoichiometry of ubiquitylation in vitro and Nedd4-mediated monoubiquitylation in cells, is not proteasome-degraded. N4BP1–3 Have Distinct Subcellular Locations—The functional interaction between N4BP1 and Nedd4 in vivo, despite the lack of co-immunoprecipitation, prompted us to determine the subcellular location of N4BP1 and Nedd4 in transfected HEK293 cells using immunofluorescence. In all transfected cells, N4BP1 was organized into discrete circular structures, which were mainly located in the nucleus and varied in number and size from cell to cell (Fig. 9, A and B). In cells with high levels of transfected protein, there was also diffuse staining in both the nucleus and cytoplasm. In contrast, Nedd4 was found mostly throughout the cytoplasm (Fig. 9, C and D), as reported previously (29). Co-transfection of the N4BP1 and Nedd4 expression vectors did not alter the subcellular localization of either protein (Fig. 9, E–G). This small degree of overlap (Fig. 9H) may underlie the failure to detect co-immunoprecipitation.

FIG. 7. In vivo polyubiquitylation of N4BP2. HEK293 cells were transfected with an expression vector for His and Xpress-tagged N4BP2, with or without Nedd4 overexpression, in the presence of Myctagged ubiquitin. N4BP2 protein was immunoprecipitated with antiHis monoclonal, and duplicate samples were analyzed. A, immunoblotting with anti-Myc monoclonal revealed many higher molecular weight species, which may represent polyubiquitylated forms of N4BP2 (lanes 1 and 2). There was no apparent enhancement with Nedd4 overexpression (lanes 3 and 4). B, equivalent expression of N4BP2 was found in each sample (lanes 1– 4), as determined by reblotting with anti-Xpress monoclonal. C, significant Nedd4 overexpression was achieved (lanes 3– 6), as determined by analyzing whole cell lysates by immunoblotting with anti-Nedd4 antiserum.

We also examined the subcellular location of transfected N4BP2 and N4BP3 in HEK293 cells. N4BP2 was found in an even distribution throughout the cytoplasm, similar to Nedd4 (Fig. 9I). However, in cells co-transfected with Nedd4, there was some redistribution of N4BP2 and Nedd4 into aggregates (Fig. 9, J–L). Transfected N4BP3 was also cytoplasmic, but was localized in vesicles of a wide range of sizes distributed throughout the cytoplasm (Fig. 9, M and N). In cells co-transfected with Nedd4, the location of Nedd4 protein was completely altered (Fig. 9O). Rather than a uniform distribution throughout the cytoplasm, all Nedd4 protein colocalized with N4BP3 in the vesicular structures (Fig. 9P). The PY mutant of N4BP3 also was found in vesicles (Fig. 9Q), but in cells cotransfected with Nedd4 the redistribution of Nedd4 was reduced significantly (Fig. 9, R–T), again demonstrating the im-

Nedd4 Developmental Targets

2905

FIG. 8. Effects of proteasome inhibition on N4BP1 and N4BP2 levels. HEK293 cells were transfected with an expression vector for flN4BP1, or expression vectors for His and Xpress-tagged N4BP2 and ␤-gal. 40 h after transfection, cells were incubated with the indicated proteasome inhibitors for 6 h. A, for flN4BP1, whole cell lysates were analyzed by immunoblotting using anti-N4BP1 antiserum. Steady state levels of flN4BP1 (band marked by asterisk) were unchanged following proteasome inhibition. B, for N4BP2 and ␤-gal, proteins were immunoprecipitated and immunoblotted with anti-Xpress monoclonal. The arrow marks the band corresponding to ␤-gal, and the asterisk marks the band for N4BP2. There was a marked increase in N4BP2 steady state levels following proteasome inhibition, but ␤-gal levels were unchanged.

portance of the PY domain for the interaction between N4BP3 and Nedd4. DISCUSSION

This study is a first step toward gaining insight into the roles played by Nedd4 in mouse embryonic development. Three novel proteins that are bona fide Nedd4 binding partners have been identified from a yeast two-hybrid screen of a mid-gestation embryo cDNA library. Two of these, N4BP1 and N4BP2, also have the capacity to be ubiquitylation substrates. For the third, N4BP3, we found no discernable evidence of ubiquitylation. However, this protein clearly associates with Nedd4 in cells, as assessed by co-immunoprecipitation and by the immunofluorescent co-localization of these proteins in vesicular structures. Two recent reports have described screens in which developmentally expressed proteins capable of physically interacting with Nedd4 were isolated (23, 28). One of these also was shown to be a likely ubiquitylation substrate (27). However, none of these were found in our screen, perhaps because of the use of different screening methodologies and/or the use of a cDNA library from a different stage of development. Somewhat surprisingly, we found that the interaction of both N4BP1 and N4BP2 with Nedd4 does not depend on proline-rich domains interacting with the WW domains of Nedd4. N4BP1 contains three proline-rich regions, but disruption of these had no effect on the in vitro binding to Nedd4. The N4BP2 fragment we analyzed lacks proline-rich domains altogether. A recent report on Rsp5, the yeast homologue of Nedd4, has shown that

its WW domains can bind target proteins in a phosphorylationdependent manner (18). It will be interesting to determine whether a similar mechanism underlies the Nedd4 interactions we have found. For both N4BP1 and N4BP2, there was a good correlation between their ubiquitylation in vitro and in cells. The in vitro ubiquitylation of N4BP1 involves either the addition of a single ubiquitin at a few sites or a single very short ubiquitin chain. In transfected cells, only monoubiquitylation of N4BP1 was found. This occurred in a Nedd4-dependent fashion and required the amino-terminal region of N4BP1; the amino-terminally truncated form of N4BP1 was not efficiently ubiquitylated in vivo, and co-expression with Nedd4 increased only flN4BP1 monoubiquitylation significantly. Only a few examples of Nedd4-mediated monoubiquitylation have been described, and all involve viral proteins (32–34). Thus, N4BP1 represents the first cellular protein identified as a Nedd4 monoubiquitylation substrate. This finding indicates that monoubiquitylation is a true cellular function of Nedd4 and not an aberrant activity co-opted by the viral replication machinery. In contrast to N4BP1, Nedd4mediated in vitro ubiquitylation of N4BP2 resulted in an extensive range of higher molecular weight forms, suggesting polyubiquitylation. Similarly, a wide range of very high molecular weight ubiquitylated forms of N4BP2 was found in transfected cells. Efficient in vivo ubiquitylation of N4BP2 was seen even in the absence of co-transfected Nedd4, suggesting that N4BP2 may be a substrate for other ubiquitin ligases in vivo,

2906

Nedd4 Developmental Targets

FIG. 9. Subcellular localization of transfected N4BP proteins. HEK293 cells were transfected with expression vectors for flN4BP1, and for His and Xpress-tagged N4BP2 and full-length human N4BP3 (KIAA0341). Indirect immunofluorescence was done for flN4BP1 using anti-N4BP1 antiserum, and for N4BP2 and N4BP3 using anti-His monoclonal. In some experiments HA-tagged Nedd4 was co-transfected and detected using either anti-Nedd4 antiserum or anti-HA monoclonal antibody. Cells were stained with 4⬘,6-diamidino-2-phenylindole to visualize nuclei. A and B, N4BP1 was found in the nucleus, mainly in discrete circular structures. C and D, Nedd4 was found in the cytoplasm. E–H, there was no apparent change in the distribution of N4BP1 or Nedd4 when cells were co-transfected with both expression vectors. As shown in H, very little overlap in distribution was apparent. A few N4BP1 dots that seem to be located outside of the nucleus were seen. I, transfected N4BP2 protein showed a cytoplasmic distribution when transfected alone. J–L, when cotransfected with Nedd4, N4BP2 formed cytoplasmic aggregates that co-localized with Nedd4. M, transfected KIAA0341 protein localized to cytoplasmic vesicles of a wide range of sizes when transfected alone. N–P, when co-transfected with Nedd4, KIAA0341 retained its distribution in vesicles. However, Nedd4 was relocalized from a general cytoplasmic distribution to the cytoplasmic vesicles containing KIAA0341. Q, transfected KIAA0341 carrying the PY domain mutation still localized to cytoplasmic vesicles. R–T, when co-transfected with Nedd4, the KIAA0341 PY mutant retained its distribution in vesicles but the amount of Nedd4 relocalized to the cytoplasmic vesicles was less than with the wild-type KIAA0341.

either in addition to or instead of Nedd4. Alternatively, perhaps only limited amounts of transfected N4BP2 are available for ubiquitylation in vivo and endogenous Nedd4 levels in these cells are sufficient. The clear functional interaction found between Nedd4 and N4BP1 was surprising given that transfected N4BP1 protein was found predominantly in the nucleus, and Nedd4 in the cytoplasm. However, it was shown recently that Nedd4 is capable of entering the nucleus but its Rev-like nuclear export sequence leads to a predominantly cytoplasmic location (35). The transient presence of Nedd4 in the nucleus may be sufficient to allow it to interact with N4BP1, as it does with the nuclear protein hPRTB, which Nedd4 regulates through ubiquitylation (35). Interestingly, hPRTB is found in discrete subnuclear structures, as is true for N4BP1. We are currently determining whether the N4BP1-containing nuclear bodies are the same as the hPRTB-containing nuclear speckles, which are implicated in transcription and RNA processing, or are related to other previously characterized classes of nuclear bodies (36, 37). Nedd4-mediated monoubiquitylation of N4BP1 may regulate N4BP1 assembly into these subnuclear structures. A recent report described the monoubiquitylation of the Fanconi anemia protein FANCD2 and showed that this modification is required for the formation of a specific class of subnuclear

structures, the ionizing radiation-inducible foci that contain Fanconi anemia protein complexes and BRCA1 (7). Although we detected no increase in the amount of transfected N4BP1 contained in nuclear bodies, or in their size or number, when monoubiquitylation was enhanced by overexpression of Nedd4, perhaps only a low level of monoubiquitylation is necessary to target N4BP1 to these structures. Alternatively, monoubiquitylation of N4BP1 may regulate its activity within these structures. In contrast to N4BP1 and N4BP2, N4BP3 is apparently not a ubiquitylation substrate. mGrb10 also has been shown to bind to Nedd4 but not be targeted for ubiquitylation (38). These proteins may serve a similar function to E6, which must physically associate with E6-AP to allow ubiquitylation of the specific target protein p53 but does not become ubiquitylated itself (12, 13). Alternatively, as has been shown for annexin XIIIb (39), the function might be to localize or enrich Nedd4 in a specific cellular compartment. N4BP3, which is assembled into vesicles of varying sizes that are distributed throughout the cytoplasm, is capable of relocating Nedd4 to these structures when the two proteins are co-expressed. Unlike annexin XIIIb, which interacts with the C2 domain of Nedd4, this relocalization is at least partially dependent on the interaction of the PY domain of N4BP3 with the WW domains of Nedd4, because

Nedd4 Developmental Targets there is some reversion to a general cytoplasmic location when Nedd4 is co-transfected with the N4BP3 PY mutant. We are currently investigating the nature of the N4BP3-containing vesicles. The presence of a consensus microsome-targeting signal (40) in the carboxyl terminus of both mouse and human N4BP3 suggests that these vesicles might be related to peroxisomes or other single-membrane organelles. Alternatively, these vesicles might be part of the endocytic system that regulates turnover of activated membrane receptors. Rsp5 has been shown to be essential in yeast for the ubiquitylation of plasma membrane proteins, leading to their internalization and targeting to the lysosome/vacuole for degradation (4, 41). A possible role for N4BP3 in mammalian cells may be to bring Nedd4 to the endocytic compartment and perhaps, by analogy to E6, target the activity of Nedd4 to an as yet undetermined substrate. In summary, we have identified three novel developmentally expressed proteins that are able to interact functionally with Nedd4. Our analysis has indicated a function for Nedd4 beyond polyubiquitylation and proteasome degradation, involving monoubiquitylation of a novel nuclear protein (N4BP1). Additionally, the PY-dependent interaction and co-localization with a second novel protein restricted to cytoplasmic vesicles (N4BP3) suggests a new function for Nedd4 in the cytoplasmic compartment as well. Future studies will address the developmental functions of these proteins and the importance of their interaction with Nedd4.

9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33.

Acknowledgments—We thank Jane P. Jensen, Alessandra Magnifico, Swati Tiwari, and Shengyun Fang for invaluable discussions and technical help, and Alfred Singer for critical reading of the manuscript.

34.

REFERENCES

35.

1. Muller, S., Hoege, C., Pyrowolakis, G., and Jentsch, S. (2001) Nat. Rev. Mol. Cell. Biol. 2, 202–210 2. Weissman, A. M. (2001) Nat. Rev. Mol. Cell. Biol. 2, 169 –178 3. Hershko, A., and Ciechanover, A. (1998) Annu. Rev. Biochem. 67, 425– 479 4. Hicke, L. (1999) Trends Cell Biol. 9, 107–112 5. Shih, S. C., Sloper-Mould, K. E., and Hicke, L. (2000) EMBO J. 19, 187–198 6. Pham, A. D., and Sauer, F. (2000) Science 289, 2357–2360 7. Garcia-Higuera, I., Taniguchi, T., Ganesan, S., Meyn, M. S., Timmers, C., Hejna, J., Grompe, M., and D’Andrea, A. D. (2001) Mol. Cell. 7, 249 –262 8. Jentsch, S., and Pyrowolakis, G. (2000) Trends Cell. Biol. 10, 335–342

36. 37. 38. 39. 40. 41.

2907

Hochstrasser, M. (2000) Nat. Cell Biol. 2, E153–E157 Ciechanover, A., Orian, A., and Schwartz, A. L. (2000) Bioessays 22, 442– 451 Joazeiro, C. A., and Weissman, A. M. (2000) Cell 102, 549 –552 Scheffner, M., Huibregtse, J. M., Vierstra, R. D., and Howley, P. M. (1993) Cell 75, 495–505 Scheffner, M., Nuber, U., and Huibregtse, J. M. (1995) Nature 373, 81– 83 Harvey, K. F., and Kumar, S. (1999) Trends Cell Biol. 9, 166 –169 Chen, H. I., and Sudol, M. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 7819 –7823 Espanel, X., and Sudol, M. (1999) J. Biol. Chem. 274, 17284 –17289 Bedford, M. T., Reed, R., and Leder, P. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 10602–10607 Lu, P. J., Zhou, X. Z., Shen, M., and Lu, K. P. (1999) Science 283, 1325–1328 Staub, O., Dho, S., Henry, P., Correa, J., Ishikawa, T., McGlade, J., and Rotin, D. (1996) EMBO J. 15, 2371–2380 Staub, O., Gautschi, I., Ishikawa, T., Breitschopf, K., Ciechanover, A., Schild, L., and Rotin, D. (1997) EMBO J. 16, 6325– 6336 Kumar, S., Tomooka, Y., and Noda, M. (1992) Biochem. Biophys. Res. Commun. 185, 1155–1161 Weston, A. D., and Underhill, T. M. (2000) Mech. Dev. 94, 247–250 Jolliffe, C. N., Harvey, K. F., Haines, B. P., Parasivam, G., and Kumar, S. (2000) Biochem. J. 351, 557–565 Zhu, H., Kavsak, P., Abdollah, S., Wrana, J. L., and Thomsen, G. H. (1999) Nature 400, 687– 693 Cornell, M., Evans, D. A., Mann, R., Fostier, M., Flasza, M., Monthatong, M., Artavanis-Tsakonas, S., and Baron, M. (1999) Genetics 152, 567–576 Qiu, L., Joazeiro, C., Fang, N., Wang, H. Y., Elly, C., Altman, Y., Fang, D., Hunter, T., and Liu, Y. C. (2000) J. Biol. Chem. 275, 35734 –35737 Pham, N., and Rotin, D. (2001) J. Biol. Chem. 276, 46995– 47003 Pham, N., Cheglakov, I., Koch, C. A., de Hoog, C. L., Moran, M. F., and Rotin, D. (2000) Curr. Biol. 10, 555–558 Hatakeyama, S., Jensen, J. P., and Weissman, A. M. (1997) J. Biol. Chem. 272, 15085–15092 Kleijnen, M. F., Shih, A. H., Zhou, P., Kumar, S., Soccio, R. E., Kedersha, N. L., Gill, G., and Howley, P. M. (2000) Mol. Cell 6, 409 – 419 Ellison, M. J., and Hochstrasser, M. (1991) J. Biol. Chem. 266, 21150 –21157 Patnaik, A., Chau, V., and Wills, J. W. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 13069 –13074 Schubert, U., Ott, D. E., Chertova, E. N., Welker, R., Tessmer, U., Princiotta, M. F., Bennink, J. R., Krausslich, H. G., and Yewdell, J. W. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 13057–13062 Ott, D. E., Coren, L. V., Chertova, E. N., Gagliardi, T. D., and Schubert, U. (2000) Virology 278, 111–121 Hamilton, M. H., Tcherepanova, I., Huibregtse, J. M., and McDonnell, D. P. (2001) J. Biol. Chem. 276, 26324 –26331 Matera, A. G. (1999) Trends Cell Biol. 9, 302–309 Zhong, S., Salomoni, P., and Pandolfi, P. P. (2000) Nat. Cell Biol. 2, E85–E90 Morrione, A., Plant, P., Valentinis, B., Staub, O., Kumar, S., Rotin, D., and Baserga, R. (1999) J. Biol. Chem. 274, 24094 –24099 Plant, P. J., Lafont, F., Lecat, S., Verkade, P., Simons, K., and Rotin, D. (2000) J. Cell Biol. 149, 1473–1484 Recalcati, S., Menotti, E., and Kuhn, L. C. (2001) J. Cell Sci. 114, 1625–1629 Rotin, D., Staub, O., and Haguenauer-Tsapis, R. (2000) J. Membr. Biol. 176, 1–17